mrfmsim custom classes

mrfmsim adds additional metadata and functionalities to the mmodel’s Node, Graph, Model, and ModelGroup classes. The inherited classes are Node, Graph, Experiment, and ExperimentGroup. In most cases these node and models are interchangeable, however it is not recommended to mix the classes in the experiment definition. For more details, see the mmodel documentation.

Node

mrfmsim.node.Node adds additional unit output in the metadata.

Graph

mrfmsim.graph.Graph changes the group “graph_module” to “mrfmsim”, and “node_type” to mrfmsim.node.Node.

Experiment

The mrfmsim.model.Experiment class is a thin wrap of the mmodel.Model class. The change is that the instantiation takes the additional dictionary input “components”, used to replace group inputs with component objects.

For example, if the model input arguments are ["magnet_radius", "magnet_origin", "field"], and a magnet component contains the attributes “radius” and “origin”, then the magnet component can be used as a replacement object with the argument:

components = {'magnet': ["magnet_radius", "magnet_origin"]}

The resulting model inputs are ["magnet", "field"]. In the event that the target name is different from the component object attribute, a custom object can be defined with the matching attribute name. The design is to allow simpler code base.

The components are not showing in the metadata output for simplicity. To inspect the components in the experiment, use the components attribute. To inspect the parameter that are replaced (subset of components) use the param_replacements attribute. Alternatively, customized metadata formatting can be used.

Component objects

An MRFM experiment typically consists of several components, such as a resonator, a magnet, and a sample. The parameters and related calculations of components can be encapsulated into an object to simplify the input process.

The ComponentBase class is an abstract dataclass with a formatted metadata output. It allows users to define the base unit of a parameter as a dataclass field. Python dataclass is used to simplify the definition process.

ExperimentGroup

The mrfmsim.group.ExperimentGroup class adds additional properties epxeriments and epxeriment_defaults alongside the parent class’s models and model_defaults properties.